Open AccessResearch on Massive News Events Evolution Prediction Based on Improved PrefixSpan Algorithm
Author(s) -
Biao Wang,
Dai Xiao,
Jing Yang,
Shun Li
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1616/1/012024
Subject(s) - pruning , event (particle physics) , sequence (biology) , computer science , set (abstract data type) , sequence database , algorithm , field (mathematics) , data mining , mathematics , biology , astrophysics , biochemistry , physics , gene , pure mathematics , agronomy , genetics , programming language
The Internet reports massive news event every day. Sequential pattern mining is adopted to study the timing relationship between news event, which can provide reference for the development or prediction of news events. With its advantages in performance and efficiency, PrefixSpan often becomes preferred algorithm in the field of sequence pattern mining. But unfortunately, because of the large amount of news and the long time span between some news events, the event sequences is long and dense, which results many subsequences in the frequent pattern and reduces the algorithm performance. In this paper, we propose an improved PrefixSpan algorithm by integrate subsequences of the supersequences in the frequent pattern and introduce news event weight in the pruning step of the PrefixSpan algorithm. Experiment results show that without affecting expression ability of frequent patterns, the improved PrefixSpan algorithm increases the efficiency by about 20 times as before and eliminates about 10% of the redundant subsequences to get a more concise frequent patterns set.